Merge pull request #35 from parkervg/feature/dialect-rework

`_dialect.py` Re-Work, `modifier` Argument for `LLMQA`, Documentation updates

KNOWN_ISSUES.MD

@@ -0,0 +1,4 @@
+```sql
+SELECT {{LLMMap('Total points?', 'w::score')}} AS Total, score FROM w 
+WHERE Total > 12
+```

README.md

@@ -134,6 +134,17 @@ SELECT COUNT(*) FROM parks
 |       1 |
 <hr>
 
+_Give me some info about the park in the state that Sarah Palin was governor of._
+```sql
+SELECT "Name", "Location", "Description" FROM parks
+  WHERE Location = {{RAGQA('Which state was Sarah Palin governor of?')}}
+```
+| Name       | Location   | Description                                                                                                                         |
+|:-----------|:-----------|:------------------------------------------------------------------------------------------------------------------------------------|
+| Everglades | Alaska     | The country's northernmost park protects an expanse of pure wilderness in Alaska's Brooks Range and has no park facilities.         |
+| Katmai     | Alaska     | This park on the Alaska Peninsula protects the Valley of Ten Thousand Smokes, an ash flow formed by the 1912 eruption of Novarupta. |
+<hr>
+
 _What's the difference in visitors for those parks with a superlative in their description vs. those without?_
 ```sql
 SELECT SUM(CAST(REPLACE("Recreation Visitors (2022)", ',', '') AS integer)) AS "Total Visitors", 

blendsql/__init__.py

@@ -1,3 +1,11 @@
 from .blend import blend
-from .ingredients import LLMQA, LLMMap, LLMJoin, LLMValidate, ImageCaption
+from .ingredients import (
+    LLMQA,
+    LLMMap,
+    LLMJoin,
+    LLMValidate,
+    ImageCaption,
+    BingWebSearch,
+    RAGQA,
+)
 from . import _configure as config

blendsql/_configure.py

@@ -4,6 +4,13 @@
 ASYNC_LIMIT_KEY = "BLENDSQL_ASYNC_LIMIT"
 DEFAULT_ASYNC_LIMIT = "10"
 
+MAX_OPTIONS_IN_PROMPT_KEY = "MAX_OPTIONS_IN_PROMPT"
+DEFAULT_MAX_OPTIONS_IN_PROMPT = 50
+
 
 def set_async_limit(n: int):
     os.environ[ASYNC_LIMIT_KEY] = str(n)
+
+
+def set_max_options_in_prompt(n: int):
+    os.environ[MAX_OPTIONS_IN_PROMPT_KEY] = str(n)

blendsql/_constants.py

@@ -1,3 +1,4 @@
+from typing import NewType, Literal, Union, Optional, Dict
 from enum import Enum, EnumMeta
 from dataclasses import dataclass
 
@@ -12,12 +13,17 @@ def __contains__(cls, item):
 DEFAULT_ANS_SEP = ";"
 DEFAULT_NAN_ANS = "-"
 
+# The 'modifier' arg can be either '*' or '+',
+#   or any string matching '{\d+}'
+ModifierType = NewType("modifier", Union[Literal["*", "+"], str, None])
+
 
 class IngredientType(str, Enum, metaclass=StrInMeta):
     MAP = "MAP"
     STRING = "STRING"
     QA = "QA"
     JOIN = "JOIN"
+    ALIAS = "ALIAS"
 
 
 @dataclass
@@ -28,3 +34,42 @@ class IngredientKwarg:
     OPTIONS: str = "options"
     REGEX: str = "regex"
     MODEL: str = "model"
+    OUTPUT_TYPE: str = "output_type"
+    EXAMPLE_OUTPUTS: str = "example_outputs"
+
+
+@dataclass
+class DataType:
+    _name: str
+    regex: Optional[str]
+    modifier: Optional[ModifierType]
+
+    @property
+    def name(self) -> str:
+        if self._name != "list" and self.modifier is not None:
+            return f"List[{self._name}]"
+        return self._name
+
+
+@dataclass
+class DataTypes:
+    BOOL = lambda modifier=None: DataType("bool", "(t|f)", modifier)
+    # SQLite max is 18446744073709551615
+    # This is 20 digits long, so to be safe, cap the generation at 18
+    INT = lambda modifier=None: DataType("int", "(\d{1,18})", modifier)
+    FLOAT = lambda modifier=None: DataType("float", "(\d(\d|\.)*)", modifier)
+    STR = lambda modifier=None: DataType("str", None, modifier)
+    LIST = lambda modifier="*": DataType("list", None, modifier)
+    ANY = lambda modifier=None: DataType("Any", None, modifier)
+
+
+STR_TO_DATATYPE: Dict[str, DataType] = {
+    "str": DataTypes.STR(),
+    "int": DataTypes.INT(),
+    "float": DataTypes.FLOAT(),
+    "bool": DataTypes.BOOL(),
+    "List[str]": DataTypes.STR("*"),
+    "List[int]": DataTypes.INT("*"),
+    "List[float]": DataTypes.FLOAT("*"),
+    "List[bool]": DataTypes.BOOL("*"),
+}